Factors associated with language concordant cognitive testing among linguistically diverse older adults

Abstract

Quality standards recommend cognitive assessments be completed in patients' preferred languages,¹ but rates of language-concordant or interpreter-mediated cognitive testing for patients who do not speak the dominant societal language are unknown.

We conducted a retrospective cohort study of patients aged 65 years and older presenting to an academic geriatric oncology clinic in Toronto, Canada who underwent cognitive testing during a comprehensive geriatric assessment between July 2015 and December 2022. A sample of 1800 patient charts was reviewed for self-reported language preference, a variable previously validated at our institution.² Demographic characteristics, patient diagnoses, reason for referral, and interpretation use were collected from the electronic medical record.

We evaluated the proportion of older patients who received language-concordant or interpreter-mediated cognitive testing, ad hoc interpretation, or no interpretation and used descriptive statistics to compare patient characteristics across two groups (language-concordant or interpreter-mediated vs. ad hoc interpretation or no interpretation). We used univariate and multivariable logistic regression to identify factors associated with cognitive testing in patients' preferred language. Statistical analyses were performed using SAS (SAS institute Inc., NC; version 9.4) and a two-sided p-value of <0.05 was considered statistically significant. The University Health Network's Research Ethics Board approved the study (see Supplemental File S1).

Of 1800 charts screened, 253 patients reported a non-English preferred language (mean age 81.2 [SD 6.4] years, 54.5% female, Table 1). The most common preferred languages were Chinese (21.3%), Portuguese (14.6%), and Italian (15.0%). Cognitive testing was language-concordant or interpreter-mediated in 45 cases (17.8%) with 49% in person, 16% by telephone, 29% unknown, and 7% documented as patient-physician shared language. For the remaining patients, 19.0% (n = 48) had ad hoc interpretation and 63.2% (n = 160) received no interpretation (Figure 1).

Patients referred for cognition (adjusted odds ratio [aOR] 8.90, 95% confidence interval [CI]: 1.76–45.09), with a curative treatment intent (aOR 3.19, 95% CI: 1.25–8.19), and who preferred a Chinese language (aOR 2.75, 95% CI: 1.02–7.36) had increased odds of language-concordant or interpreter-mediated cognitive testing, whereas patients with higher comorbidity (Charlson Comorbidity Index ≥2 vs. 0 aOR 0.17, 95% CI: 0.06–0.47), and who resided longer in Canada (aOR 0.97 per year, 95% CI: 0.95–1.00) had reduced odds (see Supplemental Table S1).

Among patients with a non-English preferred language presenting to a geriatric oncology clinic, only 17.9% received interpreter-mediated or language-concordant cognitive testing. Factors associated with increased odds of interpreter-mediated or language-concordant testing included being referred for cognition, having a curative treatment intent, and preferring a Chinese language. Patients with higher comorbidity and who lived longer in Canada had reduced odds of interpreter-mediated or language-concordant cognitive testing.

Ensuring linguistically diverse patients receive cognitive testing in their preferred language is critical, as language discordant assessments may delay dementia diagnosis and management.³ Our results are striking given our institution funds in-person, video, and telephone interpretation. This suggests other barriers, including time, ease of interpretation,⁴ and possibly, language prejudice and discrimination,⁵ must also be explored.

Our finding that patients with curative intent or who were referred for cognition had increased odds of language-concordant testing is appropriate given that the clinic often recommends alterations to treatment plans in patients with cognitive impairment.⁶ Conversely, patients with higher comorbidity had reduced odds of language-concordant testing, which may stem from decisions to de-emphasize cognition when other comorbidities provide sufficient rationale to alter treatment. The reduced odds of language-concordant testing in patients who lived longer in Canada could relate to assumptions that patients have “assimilated” the dominant societal language. Other provider assumptions about a patient's language proficiency based on race or accent⁷ might explain our finding of patients who preferred a Chinese language having increased odds of language-concordant or interpreter-mediated cognitive testing.

Limitations of our study include potential documentation inaccuracies, which may underestimate our reported interpretation rates. Additionally, we could not capture patients' English proficiency level, which might have influenced decisions to involve interpretation services. Our multivariable analysis may not have been adequately powered; therefore, predictors should be viewed as hypothesis generating. Finally, our study does not account for the complexities in behavior and associations between language and gender, race, and socioeconomic status.

Our study demonstrated inadequacy of interpreter-mediated or language-concordant cognitive testing for older patients preferring a non-English language at an academic hospital where interpreters are funded and accessible. Providing interpretation services is necessary, but likely an insufficient sole intervention to enable cognitive testing in a patient's preferred language. Further research is required to understand the facilitators and barriers of language-concordant and interpreter-mediated care, unique to the older adult population experiencing cognitive impairment, and to determine areas for systemic improvement.

V.L.C. was involved in the concept and design, acquisition of data, interpretation of data, preparation and approval of the manuscript. A.J.W. was involved in the acquisition of data, preparation and approval of the manuscript. S.M.H.A. was involved in the concept and design, interpretation of data, preparation and approval of the manuscript. S.R. was involved in the concept and design, interpretation of data, preparation and approval of the manuscript. X.Y.Y. was involved in the analysis of data, preparation and approval of the manuscript. C.R.-R. was involved in the concept and design, interpretation of data, preparation and approval of the manuscript.

A.J.W. received funding for employment from our sponsors to contribute to this research. All other authors (V.L.C., S.M.H.A., S.R., X.Y.Y., and C.R.-R.) have no conflicts of interest to declare.

The sponsors did not have any role in the study design, data analysis or interpretation, preparation, review, or the approval of the final manuscript.

This work was supported by the Savlov Family Foundation, Oskar Scher Schmidt Charitable Foundation, and Marilyn & Charles Baillie Family Foundation, through the Sinai Health Systems Geriatrics Summer Scholars Program. The funders did not have any role in the study design, data analysis or interpretation, preparation, review and approval of the final manuscript. The results from this research were presented at the 2024 Canadian Geriatrics Society Annual Scientific Meeting as an oral presentation.